Re: [Texmacs-dev] Arch mirror

[Philippe Audebaud]

Hi *,

On Wed, Oct 01, 2003 at 01:55:52PM +0200, Joris van der Hoeven wrote:

> Following the discussion about CVS/Arch: if  I use such a system, are there
> any people on this list who  would make contributions which they don't make
> now? If yes, then what kind of contributions?

1) a CVS archive is important as such, for general purposes.

   It is really important to know about changes where and when there have
   been made : both by following changes through successive commits and with
   the help of changelog.

   As a single example, at some point guile 1.6 started been accepted both
   for compilation and execution of texmacs.bin. What kind of patch has been
   required? Where? When? Submitted by who? For sure, the link
   http://www.texmacs.org/Web/Changes.html provides *no* answer.

   The same page contains 'Many bug fixes (1.0.1.23)' : and then???

   For now, versions 1.0.1.14 < _ < 1.0.1.21 + 23 vanished. And with them all
   the raw minimal chronology these sources packages provided.

   So for giving concrete examples in complement to Alvaro and David strong
   argumentations.

2) opening the system will offer the project more volonteers.

   CVS and the likes have been designed for collaborative purposes. Otherwise,
   what need, if working alone (could be RCS if any)? Even if everybody knows
   well about its imperfections, it is widely used, people are familiar with
   it and such a tool is *right now* available. 

   Besides, there exists various 'development models' there around. The Linux
   Kernel  is one which  could seem  close to  TeXmacs one : for  sure Linus
   decides at the very end. Meanwhile, the model is open enough to allow for
   many people playing with the archive, submitting patches against the
   continuously available successive archive snapshots. 

   Also 'senior' contributors work on their own series either for experimental 
   purposes (dev_fs, ide_*, etc) or anything else, then getting  their changes
   been sync'ed  with the  major branch  when  and if appropriate. 

   TeXmacs still need to be provided interfaces with major desktops. Is there
   any activity around the GTK port for instance ? While so many people are
   competent on this side, they are used at working in another model; if we
   need them, and other skills as well (that I do believe) let us provide a
   better way to letting them playing around with the sources! 

   For instance, at some point the native port to Windows has been integrated? 
   Actually, I do not know exactly. And what about the meaning of 'Further
   preparations for Windows version (1.0.1.23'? If I were about to think
   about porting to yet another environment, for sure, I'd like to follow
   which changes have been required, and why, etc? 

3) it is important for me, wrt my own developments related to TeXmacs.

   Previous arguments apply to me as well...

   My priorities and TeXmacs mainstream ones could be different. So I *need*
   to be able to provide accurate patches for the changes or experiments I do
   against TeXmacs sources. Otherwise, waiting for *my* changes to be
   integrated later, not even being sure about their status is not a solution 
   for me. Same thing apply to plugins, which could depend on the version as
   well, or other pieces of softs, etc.

   For the very moment, I manage with a local CVS archive. Fortunately,
   things became simpler when the 'TeXmacs-<version>' thing  disappeared. 

   Aother point is related to the kind of activity done on TeXmacs. This is a
   loss  of time  doing things  which are  on their  way elsewhere,  or worse
   worked  out by  Joris!.. My personal  and  general feeling  is about  the
   opacity around TeXmacs development. Sharing an archive could reduce the (my)
   trouble, by providing at least some  piece of  information.(Well, CVS is
   not enough; a good start anyway.)

To  turn your  question the  other way  round, one  could think  as  well the
choices  made for  a development  model have  (also) a  big incidence  on how
people contribute  to a project?   Compare Gnumeric and Gnucash  for instance
which are both major applications to offer average users.

-- 
Philippe Audebaud, INRIA, 2004 Route des Lucioles,
BP 93, 06902 Sophia-Antipolis Cedex, FRANCE

Tel.: +33 (0)4 9238 5051   Fax: +33 (0)4 9238 7765
http://www-sop.inria.fr/lemme/Philippe.Audebaud/